Re: [hatari-devel] FPU status register handling & fmovem.x bugs

[Andreas Grabher <andreas_g86@xxxxxxxxxx>]

Am 03.06.2018 um 16:21 schrieb Toni Wilen <twilen@xxxxxxxxxx>:

>>>>>                      moveq #-1,d0
>>>>>                      fmove.l d0,fpsr
>>>>>                      fmove.l fpsr,return_value
>>>> 
>>>> Quick 68060 result: 0FFFFFF8 (FPCR=0000FFF0)
>>>> 6888x and 68040 may return different values, I can't test until later
>>>> this weekend.
>>> 
>>> In case results on 68040 and 6888x are the same, I suggest the appended patch.
>>> <fpu_patch22_03-01-2018.diff>
>> 
>> Appended is a corrected patch without my internal changes.
> 
> Real 68882 has identical behavior. I don't have 68881.
> 68040 FPCR is different: 000000FFFF (4 undefined bits are read/write).
> FPSR is identical.

Corrected patch is appended.

Attachment: fpu_patch22b.diff
Description: Binary data